@Cavan-Ou

Battle-tested multi-agent collaboration playbook for DeepSeek Harness: model-tier routing, spec discipline, git single-writer rule — as an installable skill. 多 agent 管线里运行 DSH 的实战协作规范

Zero-touch runtime telemetry for DeepSeek Harness: every session writes its own report card (task/model/tools/failures/duration) into a human-readable journal. 纯观察者运行事实遥测。

V4 Flash "god mode" for DeepSeek Harness: w7 persona anchoring + first-turn tool anchoring + complexity-dispatched guidance (headless-native, opencode-go verified).